Vector spin chirality in classical & quantum spin systems
Jung Hoon Han (SKKU, ) Raoul Dillenschneider, Jung Hoon Kim, Jin Hong Park (SKKU) Shigeki Onoda (RIKEN) Naoto Nagaosa (U. Tokyo) Cij= Motivation:
A recent flurry of activities on the so-called multiferroic materials As a result, a new type of coupling of spin & lattice came into focus The key theoretical concept appears to be: Two noncollinear spins form a nonzero vector spin chirality (spin chirality) Cij= The spin chirality couples to the lattice displacement or dipole moment Pij Cited: 248 times as of April, 2007
Control of ferroelctricity using magnetism Nature 426, 55 (2003) Cited: 248 times as of April, 2007 TbMnO3 Mn d orbitals are orbital ordered Magnetic interaction contain both ferro and antiferro exchanges Fertile ground for frustrated magnetism Cheong & Mostovoy Review
Control of ferroelctricity using magnetic field Cheong & Mostovoy Review Nature Materials 6, 13 (2007) At low temperature, both magnetic & dipole order exist Impossible without the existence of a strong coupling between magnetic order & dipole order Magnetic field along b switches polarization from c to b axis in TbMnO3 (Kimura) Connection to Spiral Magnetism
The original work of Kimura demonstrated controllability of FE through applied field Connection to spiral magnetism made clear by later neutron scattering study of Kenzelman et al. T Theoretical idea With spiral magnetism one can define a SENSE of ROTATION (chirality) given by the outer product of two adjacent spins Cij= Cij is odd under inversion, another order parameter of the same symmetry, Pij(olarization), can couple linearly to it as Cij*Pij For collinear antiferromagnetism the sense of rotation is ill-defined INVERSION INVERSION Spin-polarization coupling
Dzyaloshinskii-Moriya type: (RMnO3 and many others) In the DM interaction, Pij is a static vector, in multiferroics, Pij is dynamic. Katsura, Nagaosa, BalatskyPRL 95, (2006) Jia, Onoda, Nagaosa, HanarXiv:cond-mat/ Multiferroics with noncollinear magnetic and ferroelectric phase
Material d-electron Polarization (C/m2); Q Specifics TbMnO3 (Kimura et al Nature 2003; Kenzelman et al PRL 2005) d4 (t2g)3 (eg)1 800; q~0.27 Orbital order Ni3V2O8 (Lawes et al PRL 2005) d8(t2g)6 (eg)2 100; q~0.27 Kagome Ba0.5Sr1.5Zn2Fe12O22 (Kimura et al PRL 2005) d5 [(t2g)3 (eg)2] 150 (B=1T); N/A N/A CoCr2O4 (Yamasaki et al. PRL 2006) Co2+ : d7 (e)4 (t2)3 Cr3+ : d3 (t2g)3 2; [qq0] q~0.63 ferrimagnetic MnWO4 (Taniguchi et al. PRL 2006) d5 (t2g)3 (eg)2 50; q=(-.214, .5, .457) CuFeO2 (Kimura et al PRB 2006) 400 (B>10T); 1/5 Can we find one? XXZ Hamiltonian with DM
Including Dzyalonshinskii-Moriya (DM) interaction DM interaction -> phase angle, flux Staggered DM, arbitrary DM
M OM O M OM O M OMO M O M Staggered oxygen shifts gives rise to staggered DM interaction -> staggered phase angle, staggered flux We can consider the most general case of arbitrary phase angles: Connecting non-chiral & chiral Hamiltonians
Define the model on a ring with N sites: Carry out unitary rotations on spins Choose angles such that This is possible provided Hamiltonian is rotated back to XXZ: Dillenschneider, Kim, Han arXiv: 0705.3993 [cond-mat.str-el]
Connecting non-chiral & chiral Hamiltonians Eigenstates are similarly connected: Dillenschneider, Kim, Han arXiv: [cond-mat.str-el] Shekhtman, Entin-Wohlman, Aharony PRL 69, 839 (1992)
DM interaction and its connection to gauge transformation was noted earlier. Shekhtman, Entin-Wohlman, Aharony PRL 69, 839 (1992) Connecting non-chiral & chiral eigenstates
Correlation functions are also connected. In particular, Since and It follows that a non-zero spin chirality must exist in Eigenstates of are generally chiral. Analogy with persistent current
For S=1/2, Jordan-Wigner mapping gives Spin chirality maps onto bond current When the flux is integral, there will be no persistent current !? The gauge-invariant definition of fermion current is Generating chiral states
Given a Hamiltonian with non-chiral eigenstates, a new Hamiltonian with chiral eigenstates will be generated with non-uniform U(1) rotations: Arovas, Auerbach, Haldane
AKLT Primer (in Schwinger bosonese) Well-known Affleck-Kennedy-Lieb-Tasaki (AKLT) ground states and parent Hamiltonians can be generalized in a similar way Using Schwinger boson singlet operators AKLT ground state is Arovas, Auerbach, Haldane PRL 60, 531 (1988) From AKLT to chiral-AKLT
Aforementioned U(1) rotations correspond to Chiral-AKLT ground state is Correlations in chiral-AKLT states
Equal-time correlations of chiral-AKLT states easily obtained as chiral rotations of known correlations of AKLT states: With AKLT: With chiral-AKLT: Excitation energies in SMA
Calculate excited state energies in single-mode approximation (SMA) for uniformly chiral AKLT state: With AKLT: With chiral-AKLT: Excitation energies in SMA Higher-dimensional chiral AKLT
M = 2S/(coordination number) For chiral Hamiltonian, replace In the original AKLT Hamiltonian (For proof, see our paper arXiv: ) In higher dimensions, it is not always possible to associate Coffey, Rice, Zhang PRB 44, 10112 (1991)
A place to look? (perhaps LaCu2O4 ) Cu Tilted up the plane Tilted down the plane Coffey, Rice, ZhangPRB 44, (1991) Cheong, Thompson, Fisk PRB 39, 4395 (1989)
A place to look? (perhaps LaCu2O4 ) Cheong, Thompson, Fisk PRB 39, 4395 (1989) Quantum mechanical spin chirality is
readily embodied by the DM interaction What is condensed by DM is more like the chiral solid, not chiral liquid Search for vector spin chiral liquid will be interesting Spin chirality in classical XY Model
The spins along a given axis rotates either clockwise (+1) or counterclockwise (-1). Ising variable can be associated. + - Magnetic order implies chirality ordering. But can chirality order first before magnetism does? If so, one has the classical chiral spin liquid phase. Classical: J1-J2 XY Model
We consider the modification of the antiferromagnetic XY model on the triangular lattice For J2=0 this is AFXY on triangular lattice (long history associated with chirality transition) Classical: J1-J2 XY Model
We consider the modification of the antiferromagnetic XY model on the triangular lattice For J1=0 this model supports nematic spin state due to equivalence of (macroscopic degeneracy) Classical: J1-J2 XY Model
(Classical) statistical properties of J2-only model is identical to those of J1-only Two types of the transition found on triangular lattice Kosterlitz-Thouless(KT) transition Chirality transition XY model on triangular lattice
Magnetic Paramagnetic TKT =0.501 T =0.512 - + Chiral The separation of the phase transition temperatures is extremely small. The chirality-ordered phase is not well-defined. Sooyeul Lee and Koo-Chul Lee, Phys. Rev. B 57, 8472 (1998) Classical: J1-J2 XY Model
We find wide temperature region where chirality is condensed,when J1 and J2 co-exist, and J2 >> J1 T Magnetic Chiral Paramagnetic J2/J1 =0 J2/J1 =9 Specific heat J2/J1= 9 (L = 15, 30, 60). T1 T2 Two phase transitions
clearly identified T1 T2 Magnetic order Nematic order Binder cumulent TKT = 0.460 Helicity modulus Helicity Modulus TKT = 0.459
This TKT must agree with the one obtained from Binder cumulent in the previous page. TKT = 0.459 . Critical phase for nematic order below TKT
disorder critical TKT We find critical dependence of N1 and N2 on the lattice dimension L below TKT. Chiral order Chiral order undergoes two phase transitions.
The first one at higher temperature obeys a scaling plot. A scaling plot of chirality using the = 0.15, = 0.69, and T = This T is higher than TKTof the nematic order. Z2 (chiral) symmetry in nematic & magnetic phase Degenerate states are chiral counterparts
Magnetic states: are opposite Nematic states: are opposite Magnetic order IMPLIES chirality symmetry breaking in Nematic order IMPLIES chirality symmetry breaking in Does nematic order IMPLY chirality symmetry breaking in ?? I can give an argument that in fact it does. In the presence of nematic ordering, each triangle supports the following configurations
+1 -1/3 Each spin orientation takes up Ising values +1 (outward) or -1 (inward) J2 interaction vanishes within the nematic manifold
J1 interaction lifts the degeneracy, results in the effective Hamiltonian valid within the nematic manifold At any finite temperature the average of the energy is negative -> the chirality must be positive!! +1 -1/3 Phase diagram: J1-J2 XY model
T paramagnetic magnetic chiral, non-magnetic J2/J1=9 Quantum version ? The nematic order already breaks the chiral symmetry and opens up the possibility of the chirality condensation well above the magnetic transition. We are trying to see if the same mechanism will result in chirality-condensed phase in quantum spin models on triangular lattice: Appendix +1 +1 J1-J2 XY Model (classical)
We consider the modification of the antiferromagnetic XY model on the triangular lattice In the spin language this is equivalent to putting a bi-quadratic interaction, and =0 in Zittartz model We did Monte Carlo (MC) on the classical J1-J2 model Nematic vs. Magnetic Ordering in Lattice
Nematic ordering Magnetic ordering Z2 (chiral) symmetry in nematic & magnetic phase Degenerate states are chiral counterparts
Magnetic states: are opposite Nematic states: are opposite Degenerate states are chiral counterparts Magnetic states: are opposite Nematic states: are opposite Order Parameters Order parameters referring to magnetic, nematic, and chiral orders are defined Magnetic, Nematic, Chiral Phase Diagram The interaction strengths are parameterized as follows
Qualitatively the phase diagram looks like X=0 X=1 T paramagnetic magnetic nematic Nematic phase also appears to be chiral Lesson? Perhaps a spin system with a sufficiently large frustrating interaction will support a chiral phase, and hence a ferroelectricity, without having the magnetic order